Performance of Fetal Medicine Foundation Software for Pre-Eclampsia Prediction Upon Marker Customization: Cross-Sectional Study

Abstract

BackgroundFMF2012 is an algorithm developed by the Fetal Medicine Foundation (FMF) to predict pre-eclampsia on the basis of maternal characteristics combined with biophysical and biochemical markers. Afro-Caribbean ethnicity is the second risk factor, in magnitude, found in populations tested by FMF, which was not confirmed in a Brazilian setting.ObjectiveThis study aimed to analyze the performance of pre-eclampsia prediction software by customization of maternal ethnicity.MethodsThis was a cross-sectional observational study, with secondary evaluation of data from FMF first trimester screening tests of singleton pregnancies. Risk scores were calculated from maternal characteristics and biophysical markers, and they were presented as the risk for early pre-eclampsia (PE34) and preterm pre-eclampsia (PE37). The following steps were followed: (1) identification of women characterized as black ethnicity; (2) calculation of early and preterm pre-eclampsia risk, reclassifying them as white, which generated a new score; (3) comparison of the proportions of women categorized as high risk between the original and new scores; (4) construction of the receiver operator characteristic curve; (5) calculation of the area under the curve, sensitivity, and false positive rate; and (6) comparison of the area under the curve, sensitivity, and false positive rate of the original with the new risk by chi-square test.ResultsA total of 1531 cases were included in the final sample, with 219 out of 1531 cases (14.30; 95% CI 12.5-16.0) and 182 out of 1531 cases (11.88%; 95% CI 10.3-13.5) classified as high risk for pre-eclampsia development, originally and after recalculating the new risk, respectively. The comparison of FMF2012 predictive model performance between the originally estimated risks and the estimated new risks showed that the difference was not significant for sensitivity and area under the curve, but it was significant for false positive rate.ConclusionsWe conclude that black ethnicity classification of Brazilian pregnant women by the FMF2012 algorithm increases the false positive rate. Suppressing ethnicity effect did not improve the test sensitivity. By modifying demographic characteristics, it is possible to improve some performance aspects of clinical prediction tests.

Nghiên Cứu Kết Quả Sàng Lọc Tiền Sản Giật Giai Đoạn Sớm Thai Kỳ Tại Bệnh Viện Phụ Sản Thành Phố Cần Thơ

“…Nghiên cứu của chúng tôi cho tỷ lệ thai phụ có nguy cơ cao mắc tiền sản giật là 10,0%, tỷ lệ này tương đương với một số nghiên cứu cùng sử dụng thuật toán trên ở cả trong và ngoài nước như: ASPRE (2017) với 10,5% [5], Rezende (2019, Brazil) với 11,9% [8], Huỳnh Thị Tuyết Mai (2021, Bệnh viện Quốc Tế Thái Hòa) với 11,3% [6] và có khác biệt lớn với Trần Mạnh Linh (2020, Huế) với 21,98% [7]. Sự khác biệt trên xảy ra có thể do trong nghiên cứu của Trần Mạnh Linh có nhiều thai phụ mang yếu tố nguy cơ hơn so với nghiên cứu của chúng tôi như: tiền sử mắc tiền sản giật, tiền sử gia đình mắc tiền sản giật, tiếp xúc thuốc lá trong khi mang thai… Từ đó có thể thấy, tỷ lệ nhóm nguy cơ cao tiền sản giật được sàng lọc thông qua thuật toán của FMF là có phụ thuộc vào đặc điểm của dân số chung về các yếu tố nguy cơ tiền sản giật, nhóm dân số càng mang nhiều yếu tố nguy cơ thì sẽ cho tỷ lệ nguy cơ cao mắc tiền sản giật cao hơn.…”

Optimising Cardiometabolic Risk Factors in Pregnancy: A Review of Risk Prediction Models Targeting Gestational Diabetes and Hypertensive Disorders

“…While other prediction models treat PE as a binary outcome, the competing risk model treats PE as an event in time, so that risk stratification can be performed at different stages of pregnancy [70][71][72]. The FMF model has undergone extensive internal and external validation in several geographically diverse cohorts [64,[73][74][75][76], with detection rates of 90% and 75% for the prediction of early and preterm preeclampsia, respectively, with a 10% false-positive rate [76]. To date, the FMF model appears to be the most well-studied [76], and external validation studies have shown comparable predictive performance corresponding to the original studies [58].…”